AI Conversation Designer

You will be responsible for developing optimal prompts and context management strategies for a variety of tasks involving large language models (LLMs) for our platform with a focus on subagents.

This includes reviewing conversations with our Superhumans and reviewing feedback in order to edit and author complex, dynamic prompts within our dialogue platform and create new agentic workflows for before, after, and during our AI agents' conversations. You will also document your work and teach others within the company. As the landscape evolves and the company grows, we expect this role to adapt accordingly.

What You’ll Do

• Design the architecture for specialized subagents operating within a live customer conversation, including technical Q&A, product and pricing expertise, discovery, objection handling, demo selection, account research, compliance, and action execution.

• Build routing and delegation systems that determine when to answer directly, invoke a subagent, run specialists in parallel, ask a clarifying question, use a tool, escalate to a human, or decline an unsafe action.

• Define contracts between the primary conversational agent and its subagents: inputs, context boundaries, structured outputs, confidence, evidence, latency budgets, cancellation behavior, and failure handling.

• Master Our Dialogue Platform: Become an expert, answer questions, and train others on prompting both within and outside of our platform.

• Train our Superhumans: Use prompting, RAG knowledge-base creation, and fine-tuning to enhance our AI capabilities.

• Prototype development: Create small-scale prototypes and AI-assisted coding to test complex prompting and data workflows.

• Multi-media prompting: Develop prompting strategies and workflows that include various types of media (images, video, etc.), with a focus on LLMs and text embedding models.

What You’ll Bring

• 3-5 years of relevant experience in AI agents, LLMs, conversation design, or related fields.

• Strong command of language: You can hear when a response sounds scripted, off-tone, or wrong.

• A record of diagnosing problems from transcripts and explaining how you knew a fix worked.

• Mastery of LLM Prompting: Strong experience with LLM prompting, particularly with OpenAI and Anthropic models. This includes hands-on experience with APIs and mastery over prompt parameters and components—not just using ChatGPT or Claude.

• Technical Floor: Comfort with quick prototyping and iteration. You can read a system prompt, work in a console, call APIs, and incorporate them into your prototypes, and you write light code (for example, Python) to get there.

• Additional Skills: Copywriting, data science, and software engineering skills are a plus.